PURPOSE: Nonepileptic seizures (NESs) are frequently mistaken for epileptic seizures (ESs). Improved detection of patients with NESs could lead to more appropriate treatment and medical cost savings. Previous studies have shown the MMPI/MMPI-2 to be a useful predictor of NES. We hypothesized that combining the MMPI-2 with a physiologic predictor of epilepsy (routine EEG; rEEG) would result in enhanced prediction of NES. METHODS: Consecutive patients undergoing CCTV-EEG monitoring underwent rEEG evaluation and completed an MMPI-2. Patients were subsequently classified as having epilepsy (n = 91) or NESs (n = 76) by using standardized criteria. Logistic regression was used to predict seizure type classification. RESULTS: Overall classification accuracy was 74% for rEEG, 71% for MMPI-2 Hs scale, and 77% for MMPI-2 Hy scale. The model that best predicted diagnosis included rEEG, MMPI-2, and number of years since the first spell, resulting in an overall classification accuracy of 86%. CONCLUSIONS: The high accuracy achieved by the model suggests that it may be useful for screening candidates for diagnostic telemetry.
PURPOSE:Nonepileptic seizures (NESs) are frequently mistaken for epilepticseizures (ESs). Improved detection of patients with NESs could lead to more appropriate treatment and medical cost savings. Previous studies have shown the MMPI/MMPI-2 to be a useful predictor of NES. We hypothesized that combining the MMPI-2 with a physiologic predictor of epilepsy (routine EEG; rEEG) would result in enhanced prediction of NES. METHODS: Consecutive patients undergoing CCTV-EEG monitoring underwent rEEG evaluation and completed an MMPI-2. Patients were subsequently classified as having epilepsy (n = 91) or NESs (n = 76) by using standardized criteria. Logistic regression was used to predict seizure type classification. RESULTS: Overall classification accuracy was 74% for rEEG, 71% for MMPI-2 Hs scale, and 77% for MMPI-2 Hy scale. The model that best predicted diagnosis included rEEG, MMPI-2, and number of years since the first spell, resulting in an overall classification accuracy of 86%. CONCLUSIONS: The high accuracy achieved by the model suggests that it may be useful for screening candidates for diagnostic telemetry.
Authors: Dona E Cragar; David T R Berry; Toufic A Fakhoury; Jean E Cibula; Frederick A Schmitt Journal: Neuropsychol Rev Date: 2002-03 Impact factor: 7.444